This invention relates to a device for on-board diagnosis, especially for a motor vehicle system which has a plurality of functional components which are networked to one another, of which at least two are control devices. Furthermore this invention relates to a corresponding method for on-board diagnosis of a system. Such on-board diagnosis can be carried out in motor vehicles, track-guided vehicles, ships, aircraft, and the like. Hereinafter the general term “vehicle” is used for this purpose.
Fault diagnosis of control devices and of the components, especially of a motor vehicle, which are controlled with them is conventionally carried out by means of self-diagnosis by the control device itself. This means that diagnosis is limited only to the control device with the respectively controlled components. If a fault is determined by self-diagnosis, a fault message is stored in the control device. Then, the fault message is retrieved in a shop using a shop tester. Since self-diagnosis relates only to one respective control device or its controlled components, component-wide or system-wide faults cannot be detected with it. But since control devices are being installed more and more in motor vehicles and are optionally networked to one another, it is increasingly important to be able to implement fault diagnosis as well in these networked systems.
The difficulty of system-wide fault diagnosis consists in that the control devices and controlled components generally originate from different manufacturers, and therefore different implementations and interpretations of specifications can occur.
A system-wide diagnosis process is known from German Published Patent Application DE 100 51 781 A1. Diagnosis takes place there in two stages, specifically component diagnosis and central system diagnosis, a system quantity and a status assigned to the system quantity being output from the component diagnosis as a result to the central system diagnosis. So-called model data of the components and of the system are stored in a function matrix. The linking of the statuses of the system quantities to operating states and the determination of faulty functions take place using a function matrix tensor. The disadvantage here is that the very extensive function matrix which would have to take into account all functional relationships of the individual components must be stored in a corresponding diagnosis device.